The broad, long-term objectives of the proposed work are to understand how proteins establish interaction specificity at the level of individual domains and to capture this in models that can be used for computational prediction and design. The problem is central to issues of human health. Improper interactions among mutated or mis-regulated proteins can lead to disease, and achieving a deeper understanding of how this occurs is important. Further, the design of novel proteins or peptides to specifically inhibit native protein complexes would provide a route to future therapies. The interaction specificity problem can be simplified by focusing on a ubiquitous yet structurally simple protein-protein interaction motif: the alpha-helical coiled coil. Coiled coils occur throughout the proteomes of all species and are associated with a wide variety of functions. Their sequence and structural properties make them tractable for computational analysis, and coiled coils are also convenient for experimental study using biophysical methods. The proposed research involves collecting systematic experimental interaction data and using it to develop and test diverse computational techniques for predicting interactions. Computational methods will also be developed for designing coiled-coil-like peptides, and experiments are proposed to characterize the designs. The specific aims of this proposal are: (1) To test and improve methods for predicting coiled-coil structures, energies and interactions, (2) To extend existing coiled-coil design capabilities to treat a broader range of targets, (3) To measure bZIP coiled-coil interactions broadly across animal species and to develop a model of how protein- protein interaction specificity can evolve, and (4) To identify interaction specificity determinants of 2- component receptor histidine kinases and manipulate these to generate heterospecific kinases. The prediction and design methods of Aims 1 and 2 will use primarily structure-based techniques. Physically motivated energy functions will be used for evaluation, and also tested in conjunction with more empirical approaches. Aim 3 involves the development of a new protein-protein interaction assay and its use to measure >10,000 possible associations. These data will be applied to analyze the evolution of bZIP transcription factor interactions and will also be valuable for testing the computational methods of Aim 1. Aim 4 tests the hypothesis that prokaryotic histidine kinases primarily homoassociate, and proposes computational and experimental techniques for identifying determinants of their interaction specificity. Together, the four aims comprise an integrated program of structural modeling, biophysical and bioinformatic analysis and experimental testing aimed at understanding coiled-coil interaction specificity. PUBLIC HEALTH RELEVANCE: Interactions between proteins are essential to the proper functioning of cells. It is important that the correct complexes form and the incorrect ones do not, even when many closely related possibilities exist. This project addresses how the sequences and structures of coiled-coil proteins determine their interaction properties, and proposes methods for predicting and modifying coiled-coil interactions using computational techniques.